QueryWeaver MCP
FalkorDB's Text2SQL server: an agent asks the database in plain language and a schema graph helps build correct SQL
- 1,094 ★ on the repository
- Last commit
Broad access
Why this level
- The agent connects a database by connection string and runs generated SQL, which can include statements that change data
- The application stores schemas and database credentials, and port 5000 is open over HTTP without a reverse proxy
Install
- Works in
Manual install
docker run -p 5000:5000 -it falkordb/queryweaverStarts the app on port 5000; MCP is served at /mcp.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent connects a database by connection string and runs generated SQL, which can include statements that change data
- The application stores schemas and database credentials, and port 5000 is open over HTTP without a reverse proxy
This is third-party code. Review the repository files before installing.
What it does
QueryWeaver turns natural-language questions into SQL and returns the query along with the result. It keeps the database schema in a FalkorDB graph: tables, columns and the links between them, so the model sees which tables to join and invents fewer nonexistent fields. The MCP part is built into the application and enabled by default. At `/mcp` it exposes four operations: `list_databases` shows connected databases, `connect_database` connects a new one, `database_schema` returns the schema, and `query_database` answers a question. The application loads schemas from PostgreSQL, MySQL and Snowflake. A client connects over HTTP with a token in the Authorization header, and the server starts with one Docker command.
Who it is for. Analysts and developers with a PostgreSQL, MySQL or Snowflake database who want answers from the data without hand-writing queries.
Good fit when
- A product manager asks the agent to pull weekly revenue from an orders database and nobody is free to write SQL
- The schema has dozens of tables and the model keeps mixing up relations: the schema graph shows how to join them
- An agent in Claude Code has to inspect the structure of a new database before refining queries
- You want a standalone Text2SQL service with a REST API and MCP for several clients at once
Not a fit when
- The data sits in a store that has no loader: PostgreSQL, MySQL and Snowflake are supported
- A database with critical data and live write access: run the generated query on a copy first
- A one-off export to a table is enough: running a separate service for one query is overkill
Example request
Connect QueryWeaver and show how many orders each region had last month, together with the generated SQLMCP
- Transport
- http
- Authentication
- API key
Environment variables
DISABLE_MCP- Set to true to turn off the built-in MCP; it is on by default
Limitations
The application needs setup: model provider keys and parameters from `.env.example` are required, and the database schema is loaded into the graph first. The AGPL-3.0 license sets conditions on network use of modified code. Port 5000 is open for HTTP and the `/mcp` token is issued inside the app, so the service should not face the internet without a reverse proxy. The MCP registry version is 0.0.11 and the operation interface may change. Answer quality depends on the chosen model and on how clear the table names are.
How to disable. Stop the container, remove the queryweaver block from the client config and revoke the token. The built-in MCP can be turned off with `DISABLE_MCP=true`.
FAQ
How do I run QueryWeaver?
Run `docker run -p 5000:5000 -it falkordb/queryweaver` and open http://localhost:5000. To configure it, pass an env file: `docker run -p 5000:5000 --env-file .env falkordb/queryweaver`, the template is in `.env.example`.
What operations does the MCP server expose?
Four: `list_databases`, `connect_database`, `database_schema` and `query_database`. All of them revolve around Text2SQL: list databases, connect, return the schema, answer a question.
How do I connect QueryWeaver as an MCP server?
In the client config add an http server at http://127.0.0.1:5000/mcp with the header `Authorization: Bearer your_token`. The token comes from the application.
Can MCP be turned off?
Yes. Pass `DISABLE_MCP=true` in the environment, for example `docker run -p 5000:5000 -it --env DISABLE_MCP=true falkordb/queryweaver`. MCP is on by default.
Does QueryWeaver have a ready Python client?
The repository has a `queryweaver` package with client and connection modules, published on PyPI under the same name. The README has no separate description of its API, see the examples in the `examples` directory.
Discussion
No questions or reviews yet.
Related
MCP serversA text-to-SQL MCP server: an agent asks a database in words, gets SQL plus a trace, and a stated refusal when data is missing.
A SQLite-based SQL engine with plugins for apps, databases and files: the anyquery mcp command exposes that data to Claude, Cursor and other MCP clients
A single-binary MCP server for MySQL, MariaDB, PostgreSQL and SQLite: read-only by default, with optional PII redaction
One MCP server for SQLite, SQL Server, PostgreSQL and MySQL: read and change data, edit schemas and export to CSV and JSON
A desktop SQL workspace with a built-in MCP server: the agent reads the schema and runs queries in the same client you already use
A lightweight database workspace with an MCP server: the agent reads the schema, runs queries and explores data in PostgreSQL, MySQL and more
README in short
The project calls itself an open Text2SQL tool with a REST API, MCP and graph-based schema understanding. It continues with Docker startup, `.env` configuration, a section on the built-in MCP with a client config example and notes on connecting external MCP servers.